Any Way to Limit 68020 Speed?

[Retro-Nerd]

The Minimig core supports Fast-IDE hardware though, nothing you'll find on vanilla Amigas. This make the loading times really fast. A plain A1200 is very slow on the IDE bus, unless you install PFS3 and add some fastram.

[limi]

A plain A1200 is very slow on the IDE bus, unless you install PFS3 and add some fastram.

Yes indeed! But who amongst us didn’t put extra RAM in our Amigas? And who would run FFS, madness I say. PFS was (and is!) the way

In any case, the main point I wanted to make is that it’s not far from the speeds of a period-accurate Amiga for the time — think a modestly upgraded A1200 or the A4000s.

I doubt there is much interest in locking the CPU to 14MHz since cycle accuracy or lower speed isn’t helping any games (as far as I know) that require the ’020.

[ericgus09]

Generally speaking as a platform, the Amiga, unlike other machines of its era, was not entirely dependent on the CPU/cpu speed so speeding up/slowing down the CPU isn't going to be as massive a noticeable impact as say with other machines (of course this depends on the task at hand, games vs say 3d rendering), the custom chips kind run along at their own pacing and implemented in an FPGA they, themselves might be somewhat running un-cycle accurate in terms of the chip bus speeds (recall the amiga had various custom chips to do various tasks, graphics, audio, memory etc) they all shared the chip bus and can and often do work independently of the CPU , that chip bus on a real machine only worked at a certain speed, now moving all that internally into an single FPGA chip, the old speed limits that used to restrict them are more or less a moot or non-existent issue now.. so that might also be playing into your "it doesn't feel period accurate" as well .. it could be all those recreated now-in-the-fpga custom chips are running too efficiently vs the original .. (give or take some variation of this concept).. but don't forget the amiga is a uniquely different design vs her contemporaries of the era and the issue you are experiencing may not be solely related to pure CPU speed (be it too fast or too slow) it could be one or more (likely multiple factors) contributing factors beyond just the CPU speed.

[rhester72]

@ericgus09 While all you say is technically true, the ridiculous Workbench execution speed is entirely the result of an uncapped CPU clock speed. The same can be seen in the Amiga libretro core. It generally doesn't affect games, and those it does can simply be reconfigured to use a 68000 (if running from floppy, at least - One-on-One via WHDLoad on an uncapped '020 is an exercise in hilarity compared to the original game).

[ericgus09]

As I said it depends on the task at hand .. and if its using the CPU more than say leaning on the custom chips .. more simply put, productivity vs games .. (though the actual line is far more blurry .. but you get the idea).. my point is the Amiga is a unique little beastie and solving the speed issue maybe more complex than simply dialing down the cpu..

[Optiroc]

If someone submits a patch that enables more correct 020 timings, it will surely be merged.

That said, a stock A1200 was a horrible computer when released. I got one in late January 1993 (hey, 30 years ago almost to the day!) and the 020 running off chip ram exclusively was dog slow. Combined with the “too little, too late” AGA chip set, it was simply not well balanced. A year later I picked up a 030/50MHz/16MB expansion which made it so much more enjoyable. So I guess I’m saying I’m really happy with the type of Amigas you can configure with the core.

[Armakuni]

If someone submits a patch that enables more correct 020 timings, it will surely be merged.

That said, a stock A1200 was a horrible computer when released. I got one in late January 1993 (hey, 30 years ago almost to the day!) and the 020 running off chip ram exclusively was dog slow. Combined with the “too little, too late” AGA chip set, it was simply not well balanced. A year later I picked up a 030/50MHz/16MB expansion which made it so much more enjoyable. So I guess I’m saying I’m really happy with the type of Amigas you can configure with the core.

Yeah it's nice to have a nippy workbench experience, I still remember the joys of running workbench from floppy on my A500

[Malor]

It’s not really 68EC020 anymore when it handles hundreds of MB with RAM.

32 bit processor = 4gb of memory... So, hundred of MB? Easy...

The 68EC020 only connected 24 address pins, so it could only physically support 16MB of RAM. AFAIK it still did the full logical decoding of 32 bits, and then just threw errors when it couldn't physically access higher RAM, so loading up a virtual EC-ish chip with more than 16 megs shouldn't cause any particular problems. It just won't throw errors above 16MB.

Motorola really charged too much for its chips. If they'd been willing to price the good chips lower, they would have sold a lot more than they did.

Yeah it's nice to have a nippy workbench experience, I still remember the joys of running workbench from floppy on my A500

Recoverable RAM disks were a big deal. The first boot took forever, while it copied itself to RAM, but then after that you could boot faster than from a hard drive. And with a mighty 2 megabytes of fast RAM, you still had quite a bit of memory free, even with 880K taken up by the Workbench image.

Hard drives were slower, but then you got to keep all your RAM, which mattered more as programs got bigger.

[limi]

Recoverable RAM disks were a big deal. The first boot took forever, while it copied itself to RAM, but then after that you could boot faster than from a hard drive. And with a mighty 2 megabytes of fast RAM, you still had quite a bit of memory free, even with 880K taken up by the Workbench image.

Indeed, that’s exactly what I did in 1988 or so with my Amiga 500 with a trapdoor 1.8MB RAM expansion for a total of 2.3MB (the one where you had to attach a cable to one of the chips, I think) — I had a set of boot floppy disks that copied Workbench, all my utilities and music players and DPaint etc to the RAD: disk, which was a bootable RAM disk that survived resetting the computer. Imagine having an SSD in a computer in the 80s, it was fast.

Then I generally left it on for days on end, it would be very fast, much faster than any hard disk available at the time. It felt like technology from the future (which the Amiga really was). As 20MB hard drives were more expensive than the computer itself at the time, this was a great way to have a super speedy setup for creative work.

Thank you for coming to my TED talk, and I think we have veered a bit from the original subject, but I am enjoying the anecdotes in this thread 😄

[dmckean]

It felt like magic and having to eventually move to a Pentium/Windows was so painful.

[ericgus09]

felt like technology from the future (which the Amiga really was)

So much of the Amiga was "from the future" it was no understatement .. and so very true .. even to this day .. looking back it still seems crazy we had this in the 80s.

[limi]

And, incredibly, the Amiga was released in the US before the Nintendo Entertainment System. Of course not the same price bracket, computer vs. console, Famicom was out in Japan, etc — but it blew my mind first time I realized what the timelines look like.

[Malor]

It felt like magic and having to eventually move to a Pentium/Windows was so painful.

I struggled with DOS for years. I was constantly looking to recreate multitasking. I used the BBS program Telemate, because it could do its own internal multitasking, so I could at least read READMEs while downloading other things. I loathed Windows 3.1, and wasn't much fond of Windows 95/98 either... too unstable. (of course, that may have been rose-colored glasses, because the Amiga wasn't exactly Mister Stability, either.)

I did some experimenting with early Linux in about 1993, and while I was able to get it running and could multitask very nicely, there was hardly any software on that system yet. I was all, "great, it runs! Now what?" I tried it every couple of years, and would go back to DOS/Windows.

I used OS/2 for awhile, but all the native software was crazy expensive, and I didn't make a lot back then, so I eventually abandoned it in favor of 95/98.

It was until Windows NT 3.51 that I finally felt like I'd come home. It still had the Windows 3.1 GUI, which... wasn't great, but it was rock solid in comparison to everything else you could buy back then. I remember saying that it felt like a tank... it took awhile to start a task, and it didn't necessarily move that fast, but it would obliterate anything you put in front of it. Eventually.

[Caldor]

Yeah... I was looking at a video of Workbench 1.3. It was 1985 and it had features that in many ways rivaled Windows 95 and made Windows 3.11 look like a joke, and Windows 3 did not release until 1990. By the time the Amiga 1200 released though, Commodore had managed to fire all of their engineers... maybe believing that they had become too big to fail or something, and failed. AGA was up against VGA and fell short.

I remember shortly before I got my Pentium 100mhz someone tried to sell me an Amiga hard drive for 100 dollars or something like that. It was 40mb. I got to test it and had not tried using a hard drive on the Amiga before. Certainly would have been nice to have much earlier, but this was at the end of the Amiga years. Good times though

[Armakuni]

It’s not really 68EC020 anymore when it handles hundreds of MB with RAM.

32 bit processor = 4gb of memory... So, hundred of MB? Easy...

The 68EC020 only connected 24 address pins, so it could only physically support 16MB of RAM. AFAIK it still did the full logical decoding of 32 bits, and then just threw errors when it couldn't physically access higher RAM, so loading up a virtual EC-ish chip with more than 16 megs shouldn't cause any particular problems. It just won't throw errors above 16MB.

Motorola really charged too much for its chips. If they'd been willing to price the good chips lower, they would have sold a lot more than they did.

Yeah it's nice to have a nippy workbench experience, I still remember the joys of running workbench from floppy on my A500

Recoverable RAM disks were a big deal. The first boot took forever, while it copied itself to RAM, but then after that you could boot faster than from a hard drive. And with a mighty 2 megabytes of fast RAM, you still had quite a bit of memory free, even with 880K taken up by the Workbench image.

Hard drives were slower, but then you got to keep all your RAM, which mattered more as programs got bigger.

The Amiga for myself at the time and the majority of my friends was purely a games machine so our workbench usage was very limited anyway

It was a very advanced machine for its time and a great piece of engineering, I don't think we have seen such a large jump in computing since a testament to the design team

[Malor]

Yeah, it was definitely ten years ahead of everyone else, and that was ten years when things were moving incredibly fast. CPUs doubled in power every 18 months to 2 years, which in a sense put it five generations in the lead.

It was a multitasking, multithreaded OS on a 7.1MHz CPU, and it was downright usable. And AmigaOS had incredible expandability built in, being the first consumer system with device drivers and callable libraries, so retrofitting hard drives and networking to a system designed for floppies was quite easy. (although TCP/IP never really became a first-class citizen; it was more like Trumpet Winsock on Win3.11.) And its system expansion bus (Zorro) was extremely well-designed, with autoconfiguration capabilities. It was pretty comparable to PCI, but way way way earlier.

The main reason it didn't do better is because Irving Gould thought personal computers were a flash in the pan, and he literally looted Commodore of all the money he could. He walked away with like thirty million dollars. Had he chosen to invest that back into his market-leading system, that might well have become thirty billion instead. I firmly believe that if Apple had owned this technology, PCs might well not have dominated in the same way. We might mostly be using many-generations-descended variants of Zorro, instead of PCIe.

[kolla]

It’s not really 68EC020 anymore when it handles hundreds of MB with RAM.

32 bit processor = 4gb of memory... So, hundred of MB? Easy...

Not with a 68EC020 that only has 24 bit address bus, it can only address 16MB. Or in case of Amiga, just 8MB due to signed address space, and pointers by design only using posivtive addresses)… and in case of real A1200 just 4MB for PCMCIA to work properly.

[Chris23235]

Yeah, it was definitely ten years ahead of everyone else, and that was ten years when things were moving incredibly fast. CPUs doubled in power every 18 months to 2 years, which in a sense put it five generations in the lead.

To be fair when the Amiga was released the 68000 mas already 6 years old and the 7MHz was a rather slow clock speed. The 68020 was already on the market for a year at this point.

[throAU]

Yeah, it was definitely ten years ahead of everyone else, and that was ten years when things were moving incredibly fast. CPUs doubled in power every 18 months to 2 years, which in a sense put it five generations in the lead.

To be fair when the Amiga was released the 68000 mas already 6 years old and the 7MHz was a rather slow clock speed. The 68020 was already on the market for a year at this point.

The 68k when it was new though was a high end workstation CPU. Getting one in the Amiga at the time was (like with the Mac) a revolution in computing power for the masses at the time.

The A1200 as above was too little too late. It should never have shipped with no fast ram, and even with fast ram, by the time the A1200 was out, 386s were common and 486s were starting to get cheap. It really should have shipped with an '030 plus fast ram at least.

[lroby74]

Please try with 2mb Chip ram, 0mb slow ram and 0mb Fast ram (like an original not expanded) A1200 and see the difference.

[Malor]

Yeah, it was definitely ten years ahead of everyone else, and that was ten years when things were moving incredibly fast. CPUs doubled in power every 18 months to 2 years, which in a sense put it five generations in the lead.

To be fair when the Amiga was released the 68000 mas already 6 years old and the 7MHz was a rather slow clock speed. The 68020 was already on the market for a year at this point.

In 1985, 7MHz was quite fast. And the project had been running for several years while they built everything, so if they had targeted the 020, it wouldn't have been out in 85. It was originally intended to be a game console, but the devs made it into a phenomenal computer instead.

edit: remember that the PC/AT was released the year prior, at 6MHz. Per Wikipedia, it was $4,000 to $6,000, and IBM's competitors were surprised at the low price and were unable to compete right away. The base Amiga 1000 was $1500, which was a screaming deal in comparison.

[Chris23235]

Yeah, it was definitely ten years ahead of everyone else, and that was ten years when things were moving incredibly fast. CPUs doubled in power every 18 months to 2 years, which in a sense put it five generations in the lead.

To be fair when the Amiga was released the 68000 mas already 6 years old and the 7MHz was a rather slow clock speed. The 68020 was already on the market for a year at this point.

The 68k when it was new though was a high end workstation CPU. Getting one in the Amiga at the time was (like with the Mac) a revolution in computing power for the masses at the time.

The A1200 as above was too little too late. It should never have shipped with no fast ram, and even with fast ram, by the time the A1200 was out, 386s were common and 486s were starting to get cheap. It really should have shipped with an '030 plus fast ram at least.

The 68000 was the CPU everybody used in 1985, don't get me wrong the Amiga 1000 was revolutionary in many ways, but the CPU was not revolutionary. It was tried and tested tech at this time. Apple had it in use since 83 at this point and Atari was already on the market with an 68000 computer and both the Macintosh and the ST clocked higher. The 68000 was so mainstream at this time that Sinclair was able to built a machine around a budget version of the 68000 a whole year before the release of the Amiga.

[Hodor]

[...]

The A1200 as above was too little too late. It should never have shipped with no fast ram, and even with fast ram, by the time the A1200 was out, 386s were common and 486s were starting to get cheap. It really should have shipped with an '030 plus fast ram at least.

The Atari Falcon 030 was released in 1992 and it was far more powerful than the A1200. A machine that could compete much better with the PCs of that era despite the lack of success and low sale numbers. A pity, indeed.

[Malor]

The Falcon had at least one major design problem, as I recall: the DSP chip was totally bottlenecked on the system DRAM. It was a good idea, and could have been kind of amazing, but the chip was fairly crippled by the slow RAM. It probably needed a local fast RAM store to really work well, like a modern GPU's video RAM.

I've never heard this directly from any source, but I suspect the 030's performance would have been very bad with the DSP really hammering on the RAM, too. It would be rather like the Amiga's original 68000 trying to run out of CHIP RAM... the chipset took priority, and running 640x400 with 16 colors would cut your CPU performance in about half.

Chris23235: the reason the Amiga's CPU was slower than the STs or Mac's was because it was using a multiple of the NTSC timing crystal, as the system was strongly intended for use with the old standard def video signals. The actual performance of the computer was way higher than you might think from the slow clock rate, because so many tasks could be handled by the chipset. Macs and STs had to use the 68000 for everything, where the Amiga could delegate a lot of work to the chipset.

[Chris23235]

Chris23235: the reason the Amiga's CPU was slower than the STs or Mac's was because it was using a multiple of the NTSC timing crystal, as the system was strongly intended for use with the old standard def video signals. The actual performance of the computer was way higher than you might think from the slow clock rate, because so many tasks could be handled by the chipset. Macs and STs had to use the 68000 for everything, where the Amiga could delegate a lot of work to the chipset.

As I said the Amiga was revolutionary in many ways. Jay Miner evolved the concepts that made the Atari 800 such a great computer. The chipset was very powerful. But when it came to raw computing the 7 MHz were slower then the 8 of the ST. Did this matter at this time? Not really. My comment was only in regards to your remark that the use of the 68000 put the Amiga five generationshead of the competition. The 68000 was the natural choice when developing a machine that should launch in 1985. It was a great CPU but it wasn't something that put you in the lead any more.
The Amiga was a great machine, most probably the best thing you could buy when interested in doing graphics at this time, but it was not because of the CPU.
The 7 MHz did matter when it came to using the Workbench. The Amiga OS was extremely ambitious with it's multi-tasking and the UI was a real looker with all the colours but coupled with the rather limited (and slow) ammount of RAM it could become dragging to use. I had an A1000 back then and remember that switching to the ST (because of it's better text processing due to the b/w monitor), I found GEM much more limited then the Workbench, but much faster in use. The 7 MHz CPU was not the only reason for this, but it made the overall performance of the original Workbench a fair bit slower.

[Malor]

My comment was only in regards to your remark that the use of the 68000 put the Amiga five generationshead of the competition.

It was the multitasking OS that put it five generations ahead, not the CPU. It had device drivers, printer drivers (although the whole print subsystem was terrible), callable system libraries, pluggable filesystems, and an extremely sophisticated expansion bus. Retrofitting TCP/IP was easy, too, although it was never fully integrated into the OS while Commodore was in charge. I'm not sure if later AmigaOSes ever did make it a first-class part of the OS, I haven't used those much.

Those things were what made the Amiga amazing. I was endlessly irritated by the limitations of DOS and Windows in comparison, up until NT, when Microsoft finally caught up. That was ten years later, which is why I say the Amiga was five generations ahead.

edit: and, yes, the ST's mono monitor was awesome. It was very tiny, but wow it looked fantastic. That was a great reason to buy one.

[Chris23235]

My comment was only in regards to your remark that the use of the 68000 put the Amiga five generationshead of the competition.

No, it was the multitasking OS that put it five generations ahead.

Here I completely agree, multitasking on the Lisa was much more limited for sure and even if Unix had multitasking, a GUI with windows definitely was a gamechanger for multitasking.

The whole concept of the Amiga OS was definitely ahead of its time.

Edit: I never worked with an Amiga with 4MB and a HDD. I imagine this would have been something else entirely then the A1000 with one Floppy.

[throAU]

To be fair when the Amiga was released the 68000 mas already 6 years old and the 7MHz was a rather slow clock speed. The 68020 was already on the market for a year at this point.

The 68k when it was new though was a high end workstation CPU. Getting one in the Amiga at the time was (like with the Mac) a revolution in computing power for the masses at the time.

The A1200 as above was too little too late. It should never have shipped with no fast ram, and even with fast ram, by the time the A1200 was out, 386s were common and 486s were starting to get cheap. It really should have shipped with an '030 plus fast ram at least.

The 68000 was the CPU everybody used in 1985, don't get me wrong the Amiga 1000 was revolutionary in many ways, but the CPU was not revolutionary. It was tried and tested tech at this time. Apple had it in use since 83 at this point and Atari was already on the market with an 68000 computer and both the Macintosh and the ST clocked higher. The 68000 was so mainstream at this time that Sinclair was able to built a machine around a budget version of the 68000 a whole year before the release of the Amiga.

Bolded above: no, it wasn't really. The c64 was only released in 1982 and they were just hitting their stride in the mid 80s.

The mid 80s was the era of the 8 bit machines, yes the 68k existed in the very late 70s but it was by no means a mainstream affordable desktop CPU until the Amiga, ST and to a latter degree, the original Mac (which was much more expensive than both). Prior to the early-mid 80s the 68k was found in Sun workstations, and not much else.

e.g., typical early 80s 68000 based platform - $9k US with NO GRAPHICS only 3 years prior:
https://en.wikipedia.org/wiki/Sun-1

The Amiga, ST and Mac were a revolution as I originally stated above, in bringing the power of 68k to the consumer market.

edit:
lol, the Sinclair QL was a 68008 - not a 68000... it also flopped... because it was trash.

[Chris23235]

Bolded above: no, it wasn't really. The c64 was only released in 1982 and they were just hitting their stride in the mid 80s.

The mid 80s was the era of the 8 bit machines, yes the 68k existed in the very late 70s but it was by no means a mainstream affordable desktop CPU until the Amiga, ST and to a latter degree, the original Mac (which was much more expensive than both). Prior to the early-mid 80s the 68k was found in Sun workstations, and not much else.

e.g., typical early 80s 68000 based platform - $9k US with NO GRAPHICS only 3 years prior:
https://en.wikipedia.org/wiki/Sun-1

The Amiga, ST and Mac were a revolution as I originally stated above, in bringing the power of 68k to the consumer market.

edit:
lol, the Sinclair QL was a 68008 - not a 68000... it also flopped... because it was trash.

Yes the Amiga, ST, Mac and the QL brought 16-Bit technology to the masses, but Commodore was in 1985 late to the party. This was all I said. Apple hat the 68000 in use in 2 different machines at this point and Atari was already on the market.

The 68008 was a budget version of the 68000 just as I wrote, the difference lay in speed and the ability to only address 1MB and not 4MB something that was not of particular interest for a homecomputer in 1984 . Nonetheless the 68008 was part of the 680x0 family.

When the 68020 launched in 1984, the introductionary price for this chip was $487 while the 68000 was down to $15. At this point the 68000 was mainstream and not revolutionary.

https://www.wikiwand.com/en/Motorola_68020

The Sun workstations were much more then the 68000, they outperformed any Mac, Amiga and ST without any problem. In 1985 Sun immediately switched to the 68020 because the 68000 didn't offer enough performance even if they were higher clocked then the competition in 1982 already.
I also don't know what you mean by "$9k US with NO GRAPHICS only 3 years prior" when the SUN-1 in 1982 had graphic capabilities way beyond everything that an Amiga could do 3 years later.

When you want to know what was revolutionary in 1985 take a look at the SUN-3, this machine showed what was possible:

https://www.rcsri.org/library/80s/Sun-P ... t-1985.pdf

Again the Amiga was a revolutionary machine in 1985, but not because of its 6 year old CPU that was in other homecomputers already.

[throAU]

I said revolutionary for the CONSUMER MARKET. Which they were, for the price. They were not Sun or Mac price. And the QL was crippled massively with the 8 bit bus on the 68008.